Prevalence of visual impairment by age group: the Salisbury Eye Evaluationin Nursing Home Groups (SEEING) study.
Friedman DS, West SK, Munoz B, Park W, Deremeik J, Massof R, Frick K, Broman A, McGill W, Gilbert D, German P. Racial Variations in Causes of Vision Loss in Nursing HomesThe Salisbury Eye Evaluation in Nursing Home Groups (SEEING) Study. Arch Ophthalmol. 2004;122(7):1019-1024. doi:10.1001/archopht.122.7.1019
Copyright 2004 American Medical Association. All Rights Reserved.Applicable FARS/DFARS Restrictions Apply to Government Use.2004
To determine the prevalence and causes of low vision in a large sampleof nursing home residents.
Twenty-eight nursing homes on the Eastern Shore of Maryland and Delawarewere enrolled in a clinical trial to assess the impact of vision restoration/rehabilitationon nursing home residents. Visual acuity was measured using both recognitioncharts and preferential looking techniques. An ophthalmologist examined allresidents with visual acuity worse than 20/40 in the better-seeing eye anddetermined the primary cause for decreased vision. Results are reported forthe better-seeing eye.
Of 2544 eligible residents, 1591 (63%) participated, but 286 residentswere unable to respond to visual acuity testing. Of the remaining 1307 residents,496 (37%) had best-corrected visual acuity worse than 20/40 in the better-seeingeye. Causes were ascribed for 412 subjects. Rates of low vision were similarbetween African American subjects and white subjects (39% and 38%, respectively;age-adjusted P = .18). Cataract was the leading causeof low vision, responsible for 37% of low vision among white subjects and54% of low vision among African American subjects. Macular degeneration wasresponsible for 29% of low vision among white subjects but only 7% among AfricanAmerican subjects. Glaucoma caused low vision in 4% of white subjects and10% of African American subjects. Refractive error was not a frequent causeof low vision in nursing home residents.
Low vision is highly prevalent among nursing home residents, with 37%having visual acuity worse than 20/40 in the better-seeing eye. Differencesin causes of low vision between African American subjects and white subjectswere noted, with African American subjects more likely to have vision losson the basis of cataract, a readily treated condition. Appropriate interventionsfor nursing home residents, who face significant obstacles in accessing eyecare services, have the potential to improve the quality of life of this at-riskolder population.
Rates of blindness and visual impairment increase with age, as demonstratedin several population-based studies of community-dwelling older persons.1- 4 Studiesof visual impairment among nursing home residents indicate that rates arelikely even higher for them, even when matching for age.1,4- 12 Onepossible explanation for this disparity is the negative impact of vision losson independent function, resulting in more visually impaired individuals beingadmitted to nursing homes. Findings from 2 studies support this hypothesis(S.K.W., oral communication, May 2003).4 Analternative explanation is that residents of nursing homes are less likelyto obtain vision-restoring services than community-dwelling individuals, resultingin a higher prevalence of untreated eye disease. We recently reported an associationbetween visual impairment and length of stay (LOS) in the nursing home, furthersupporting this hypothesis.13
Given the high prevalence of visual impairment among nursing home residents,we have undertaken a randomized clinical trial of aggressive vision restorationvs usual care to determine whether or not such programs improve the qualityof life and function of nursing home residents.13 Wepresent the causes of vision loss among this study population.
A detailed description of the study has been published.13 Inbrief, 28 nursing homes on the Eastern Shore of Maryland and Delaware, representingall but 2 of the nursing homes on the Eastern Shore within a 2-hour driveof Salisbury, Md, at the time of study design, were enrolled in the study.Two additional nursing homes refused.
To be enrolled in this clinical trial, residents had to be 65 yearsor older and had to reside in the nursing home on a permanent basis (patientsto be discharged within 30 days were therefore excluded). Because we weremonitoring change during the course of 1 year, we also excluded those whowere known to be at risk for imminent death. Those patients who were extremelymentally compromised were also excluded at the outset because vision testingwas not possible. These residents had to be unresponsive to all stimuli, includingverbal and motion, on 2 separate occasions to be excluded on this basis. Wealso excluded those who could not provide informed consent and for whom nolegal guardian could be found and 2 patients with whom language barriers werelarge enough to preclude accurate assessment of the resident.
Habitual and best-corrected acuity was attempted on all eligible residents;we used standard letter/symbol charts and grating acuity charts, as describedbelow. The habitual, distance, refractive correction was measured with a lensometer(Nikon model OL55; Nikon Instruments, Torrence, Calif) on the participant'susual distance glasses (reading-only glasses were not used). A portable handheldauto refractor (Retinomax model 30965; Nikon Instruments) was used to obtaina starting correction if the patient's usual glasses were not available.
Initial visual acuity, using the patient's usual glasses, was testedfirst, using Early Treatment Diabetic Retinopathy Study (ETDRS) charts orLea symbols to determine if refraction was necessary. Each eye was testedseparately. The maximum number of letters read correctly was scored, and acuityin logMAR units was reported. If the visual acuity was 20/40 or better ineither eye, no further testing using a different correction was done.
If habitual visual acuity was worse than 20/40 in the better-seeingeye, a subjective refraction was attempted with a trial lens set. The startingprescription was based on the lensometer results or, if the participant didnot have glasses, on results from the autorefractor. If necessary, a streakretinoscopic evaluation of the likely refraction was also obtained. All subjectswith visual acuity worse than 20/40 in the better-seeing eye underwent anexamination by an ophthalmologist. Vision was also tested using a gratingacuity chart (Teller acuity full system; Visitech Consultants, Dayton, Ohio),following the procedures outlined by Marx et al13 fornoncommunicative older persons.14,15 Proceduresare outlined in detail in a previous publication15 andsummarized here. Lighting on the screen was standardized within 400 to 500lux, and the distance from the eye to the screen was set at 84 cm. Residentswere requested to look at a large rectangular card, which had 2 halves; oneside was blank and the other had stripes, or line gratings, of decreasingsize. The residents were asked to look at the stripes on the card. When thetechnician was satisfied that the patient saw the correct side, a card witha set of smaller gratings was presented. The subject had to correctly identifythe side with the gratings on 2 of 3 presentations to have "seen" the card.The technician started with 2.3 cycles/cm until the last card with which thesubject saw 2 of 3 presentations.
Different technicians assessed the visual acuity using the standardETDRS chart and the Teller Acuity cards, and they were unaware of the visualacuity obtained with the other test. Both initial and (if visual acuity wasworse than 20/40 in both eyes) best-corrected grating acuity were obtained.The refraction used was the same as that obtained for standard letter testing.The study protocol stated that at least 2 hours should separate the acuityascertainment using the 2 tests to avoid tiring the patients, unless the technicianfelt the patient could proceed in a shorter period of time.
For the purposes of this study, visual impairment was defined as visualacuity in the better eye worse than 20/40 on ETDRS and grating acuity. Inthe event that only 1 test could be performed, the results of that test wereused to assign patients. If there was disagreement of more than 1 line onthe ETDRS between the 2 tests, or if 1 test indicated the patient saw 20/40or better and the other indicated visual acuity worse than 20/40, a retestwas done and the results were used to assign patients. We have shown thatagreement between the 2 measures of visual acuity is good with evidence thatacuity tested with Teller cards is better when cognition is poor.15 When a discrepancy persisted, we relied on the examiningphysician to determine if the clinical findings were consistent with visualacuity worse than 20/40 in both eyes.
Blindness was defined using both the United States legal definitionof visual acuity less than or equal to 20/200 in the better-seeing eye andthe World Health Organization definition of visual acuity less than 20/400in the better-seeing eye. A board-certified ophthalmologist (D.S.F.) determinedthe primary cause of blindness at the time of the definitive examination.When more than 1 cause of vision loss was present, causes of permanent visionloss (glaucoma, age-related macular degeneration [AMD], and so on) were assignedas the primary cause if other causes (refraction and cataract) were also present.
Residents were given the Mini-Mental State Examination to determinethe level of cognitive impairment. For this test, scores range from 0 to 30,with higher scores indicating higher cognitive function. Scores from 18 to24 are consistent with moderate cognitive impairment, and scores below 10indicate severe cognitive loss. These data were collected at the time of visualacuity determination. If a resident could not complete parts of the Mini-MentalState Examination for physical reasons, the questions were not included inthe calculation of the score, which was calculated on the basis of the remainingquestions and rescaled to the reference total of 30 (for example, those itemswere skipped for residents who could not hold a pencil and write or draw).
Age, sex, race, educational status, and length of time in the nursinghome were ascertained from the medical record.
For this article, we present the cause-specific rate of visual impairmentstratified by age, group, and race. We compare the causes of visual impairmentby age, race, sex, and LOS. Logistic models accounting for correlation withinnursing homes were used to compare cause-specific rates. All procedures andprotocols for this study were reviewed and approved by the Johns Hopkins institutionalreview board, Baltimore, Md, in accordance with the Declaration of Helsinki.
A total of 3201 patients were identified in the census of the 28 nursinghomes. Of these patients, 657 were not eligible, primarily because they weretoo young (37%) or were short-stay residents (35%). However, 19% were tooseverely cognitively impaired to be in the study; this was determined at theoutset. Of the 1307 participants whose visual acuity could be tested, 496(38%) were initially found to have best-corrected visual acuity worse than20/40 in the better-seeing eye (Figure 1).Rates were similar comparing African American subjects with white subjects(39% and 38%, respectively; age-adjusted P = .18).Fifty-three subjects did not receive an eye examination (40 refused, and 13had died or were hospitalized at the time of examination), and an additional31 were found to have no ocular cause of decreased vision (vision loss wasmost probably due to poor cognition and inability to fully cooperate withtesting, although central causes of decreased vision could not be ruled out).Assuming that these individuals in fact had normal vision and that a similarproportion of those not receiving an eye examination also had normal visiondecreases the overall prevalence of low vision to 36%.
Table 1 presents the cause-specificprevalence of visual impairment and blindness by race. Cataractwas the leading cause of low vision for both races but was a more common causeof vision loss among African American subjects (24.2% of AfricanAmerican subjects vs 11.8% of white subjects; age-adjusted P = .001).Macular degeneration was uncommonamong African American subjects (2.3%), whereas it was the second leadingcause of low vision among white subjects (10.3%; age-adjusted P = .001). Glaucoma was a more common cause of low visionamong African American subjects than among white subjects (3.3% and 1.5%,respectively; age- and cluster-adjusted P = .08).We did not attempt to confirm glaucoma with visual fields, but glaucoma wasassigned as a primary cause of low vision only in cases of severe optic nervehead excavation. The prevalence of refractive error resulting in low visionwas 3% among African American subjects and 3.5% among white subjects.
Table 2 presents the cause-specificprevalence of visual impairment by age groups (<85 vs ≥85 years). Cataractwas a leading cause of low vision in all age ranges and increased with age,whereas AMD was less common in the youngest ages but explained the majorityof cases of low vision among the oldest white residents (3.2% vs 14.1%, with22% of all white subjects in this age category having AMD as a cause of lowvision). Similarly, glaucoma prevalence as a cause of low vision increaseddramatically with age (1% vs 2.9%). No association was seen between sex andspecific causes of low vision or the prevalence of low vision (age-adjusted P = .8).
Length of stay in the nursing home was associated withthe presence of low vision. On average, individuals with low vision were residentfor 37.3 months, whereas those without visual impairment were resident for24.4 months. The association with LOS remained positive after adjusting forage, sex, race, and correlation within homes (P<.001). However, no specific cause of low vision was associatedwith LOS. Furthermore, no significant difference in LOS was seen when comparingconditions for which vision can be restored (cataract and refractive error)with those for which no vision-restorative treatment is available (AMD, glaucoma,and others).
Bilateral blindness using a cutoff of visual acuity (20/200 in the better-seeingeye) occurred in 7.4% of those 85 years or older vs 3.1% in the younger subjects(P = .002). African American subjects were more likelyto be bilaterally blind than white subjects (6.5% vs 4.8%; age-adjusted P = .19), but this was not statistically significant. Age-relatedmacular degeneration accounted for 23 of 67 cases of bilateral blindness,only 1 of which occurred in an African American individual. Glaucoma accountedfor 11 cases, and cataract accounted for an additional 10.
Habitual visual acuity less than 20/40 is common in nursing homes, withalmost 40% of those studied in 28 nursing homes on the Eastern Shore of Marylandand Delaware meeting this standard of visual impairment. Contrary to findingsin community-based studies, refractive error among this population is responsiblefor a small proportion (<10%) of visual impairment. More than 5% of nursinghome residents with testable vision were bilaterally blind. Our results aresimilar to those of previous research documenting higher rates of visual impairmentand blindness among nursing home residents than among community-dwelling personsof similar age.4,6,10
African American subjects in the present study had a slightly higherprevalence of blindness, but this was not statistically significant. Thisfinding differs from a previous study on racial differences in visual impairmentamong nursing home residents in which African American persons were foundto have substantially more blindness (21% vs 14% prevalence among white persons; P<.01).6 Explanationsfor the difference in findings include secular trends (the previous studywas conducted more than a decade ago), differences in the populations studied,and bias introduced by study methodology. It is possible that access to oruse of cataract surgery services for African American persons has increasedduring the past decade, which would have resulted in a reduction in cataractblindness among African American persons as compared with past studies. Analternative explanation is that by excluding those in whom vision could notbe tested, the present study underestimated the prevalence of blindness amongAfrican American subjects. It is possible that some with testable vision whowere blind may have been misclassified, which could have resulted in lowerestimates of blindness prevalence in white subjects, African American subjects,or both. If we assume that all untestable individuals (who were excluded fromthe present study) actually had low vision, then the overall prevalence wouldhave been 48.9% for white subjects and 49.5% for African American subjects(P>.2). We therefore believe it is highly unlikelythat our exclusion criteria caused the rates to be artificially similar betweenAfrican American and white subjects. A final possibility is that in the ruralsetting of the Eastern Shore, admission to nursing homes for vision loss maybe less likely among African American persons than among white persons, leadingto a lower overall prevalence of blindness among African American nursinghome residents than was seen in an inner-city population. Whereas our experienceobserving community-dwelling individuals on the Eastern Shore found visionto be a predictor of nursing home admission, we did not find differentialadmissions to nursing homes among the visually impaired by race.
The higher prevalence of cataract among African American persons canbe attributed to any of several factors. Previous research, including ourreport on residents of the Eastern Shore, indicates that African Americanpersons are less likely to visit an ophthalmologist,16- 18 andthey obtain cataract surgery at lower rates than white persons.19,20 Therefore,African American persons with visual impairment from cataract may be morelikely to enter a nursing home than white persons. Alternatively, AfricanAmerican persons may be less likely to obtain cataract surgery services onceresiding in the nursing home. This is highly unlikely to lead to the disparitywe observed because we found that even when residents were screened for cataractand referred for surgery, fewer than 5% underwent cataract surgery duringa 1-year period (S.K.W., oral communication, May 2003).
Macular degeneration was the leading cause of blindness among whitesubjects and the second leading cause of visual impairment, whereas it explainedonly 1 case of bilateral blindness among African American subjects and 6 casesof visual impairment. This is expected, with previous research indicatingthat AMD is less common among African American persons than white persons.21- 23 Glaucoma was a relativelyuncommon cause of low vision, but this is not surprising given that glaucomararely leads to reduced central visual acuity until late in the disease.Glaucoma blindness was about 4 times as common in African Americansubjects as in white subjects, a finding that is consistent withprevious publications.22
Length of stay was associated with low vision but not with specificconditions. This increasing visual impairment with increasing LOS remainedwhen we adjusted for age, race, sex, and Mini-Mental State Examination score.This finding raises the concern that nursing home residents are not screenedfor vision loss at admission and are not adequately observed for progressionof ocular disease during their residence in the home. Incident visual impairmentand worsening of preexisting conditions are likely not being treated, resultingin higher rates of visual impairment with greater LOS. We did not find a significantassociation between treatable causes of visual impairment and LOS; however,this may be owing to the relatively small sample available for this comparison.
Our findings of high rates of visual impairment and increased prevalenceof visual impairment with greater LOS point to the need for greater emphasison vision in nursing homes. Vision screening on admission and periodic examinationscould be implemented to help identify individuals with eye diseases requiringtreatment as well as those with uncorrected refractive errors.
A substantial proportion of the residents whom we were planning to screenrefused to participate (32%). Some of these individuals would have been untestableand therefore ineligible for this study. The nursing staff estimated that12% would have been ineligible because of severe cognitive loss (as definedby no response to external stimuli on 2 or more occasions). Had we excludedthem as ineligible at the outset, our response rate would have been 71%. Therewere no age or sex differences in participation for the study; however, AfricanAmerican persons were more likely to participate than white persons. Our findingof no difference in race-specific causes of visual impairment and blindnessbetween African American subjects and white subjects could have been influencedby nonparticipation. If African American persons with vision loss were lesslikely to participate than white persons with vision loss, our estimates ofvisual impairment rates may have been lower for African American subjectsthan white subjects on this basis, thereby making rates appear equal. We donot believe that this is the case, however. In a previous publication,24 we documented that community-dwelling nonparticipantsreported similar rates of vision impairment to those of participants in theSalisbury Eye Evaluation project. Furthermore, family members made the vastmajority of refusals. We doubt that refusals by family members were made onthe basis of visual function.
Although the study enrolled nearly all nursing homes on the EasternShore of Maryland and Delaware and therefore is likely representative of visualimpairment in this region, other parts of the United States may have differentnursing home populations. Another limitation is that we were unable to testvision in a substantial proportion of nursing home residents. This contrastswith the Baltimore Nursing Home Study,6 inwhich higher testability was noted. This difference can be attributed to differencesin methodology between the studies or possibly to an increased prevalenceof severe cognitive impairment among nursing home residents during the lastdecade. It is possible that knowledge of the race of the participant influencedthe ophthalmologist in selecting diseases as causes of blindness. Given thenursing home setting and the limitations of our ability to formally documentAMD and glaucoma with photographs, all diagnoses were based on the clinicalexamination and therefore could not be independently validated. Nevertheless,the findings are similar to population-based studies that have used more objectivemeans, masked to racial characteristics, to determine disease presence. Finally,we may have incorrectly attributed vision loss to cognitive difficulties in31 subjects because no clearly identifiable cause of vision loss was present,and the subjects were unable to describe the quality of their vision. Subtlemacular lesions or occipital causes of vision loss cannot be ruled out completelyin these cases.
In summary, low vision and blindness are highly prevalentamong residents of nursing homes, with more than one third affected. AfricanAmerican subjects have a greater burden of unoperated cataract and glaucoma,while white subjects have much higher prevalence rates of AMD. Cataract remainsthe most common cause of low vision for both racial groups, pointingto a potential benefit of intervention programs aimed at providing surgicalservices to this population.
Correspondence: David S. Friedman, MD, MPH, Wilmer Eye Institute,Room 120, 600 N Wolfe St, Baltimore, MD 21287 (email@example.com).
Submitted for publication May 20, 2003; final revision received January2, 2004; accepted January 15, 2004.
This study was supported by grant AG 15812 from the National Instituteon Aging, Bethesda, Md; a Robert E. McCormick Scholarship from Research toPrevent Blindness, New York, NY (Dr Friedman); and a Dennis Jahnigen ScholarsAward from the American Geriatrics Society, New York, NY. Dr West is a seniorscientific investigator for Research to Prevent Blindness. We thank KarenBandeen-Roche, PhD, for her assistance in the design and analysis of thisresearch. We are grateful to the nursing home administrators and nurses andthe residents and their families who made this project possible.